
# installing and importing libraries and modules we can predict we will need
# more modules could be installed later
# pip install seaborn
# pip install scipy
# pip install scikit-learn

import warnings
warnings.filterwarnings('ignore')

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt 

from scipy import stats

from sklearn import linear_model
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split

from sklearn.model_selection import cross_val_score, KFold
from sklearn.linear_model import Ridge
from sklearn.linear_model import Lasso


# load data
life = pd.read_csv('Life-Expectancy-Data-Updated.csv', sep=',')
life


# explore data structure
life.info()
print('Duplicates:',life.duplicated().sum())
print('Missing values:', (life.isnull().sum().sum()))

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2864 entries, 0 to 2863
Data columns (total 21 columns):
 #   Column                       Non-Null Count  Dtype  
---  ------                       --------------  -----  
 0   Country                      2864 non-null   object 
 1   Region                       2864 non-null   object 
 2   Year                         2864 non-null   int64  
 3   Infant_deaths                2864 non-null   float64
 4   Under_five_deaths            2864 non-null   float64
 5   Adult_mortality              2864 non-null   float64
 6   Alcohol_consumption          2864 non-null   float64
 7   Hepatitis_B                  2864 non-null   int64  
 8   Measles                      2864 non-null   int64  
 9   BMI                          2864 non-null   float64
 10  Polio                        2864 non-null   int64  
 11  Diphtheria                   2864 non-null   int64  
 12  Incidents_HIV                2864 non-null   float64
 13  GDP_per_capita               2864 non-null   int64  
 14  Population_mln               2864 non-null   float64
 15  Thinness_ten_nineteen_years  2864 non-null   float64
 16  Thinness_five_nine_years     2864 non-null   float64
 17  Schooling                    2864 non-null   float64
 18  Economy_status_Developed     2864 non-null   int64  
 19  Economy_status_Developing    2864 non-null   int64  
 20  Life_expectancy              2864 non-null   float64
dtypes: float64(11), int64(8), object(2)
memory usage: 470.0+ KB
Duplicates: 0
Missing values: 0


# generate descriptive statistics
life.describe().T


life_dropped = life.copy()
life_dropped.drop(['Country', 'Region', 'Year', 'Infant_deaths', 'Under_five_deaths', 'Adult_mortality', 'Economy_status_Developed', 'Economy_status_Developing'], axis=1, inplace=True)
life_dropped


# measure the skewness of the Data
# чем дальше показатель асимметрии (skewness) от нуля, тем ассиметрия "тяжелее"
life_dropped.skew()

Alcohol_consumption            0.571192
Hepatitis_B                   -1.658973
Measles                       -0.994008
BMI                           -0.121106
Polio                         -1.726452
Diphtheria                    -1.826928
Incidents_HIV                  4.978735
GDP_per_capita                 2.377248
Population_mln                 8.157684
Thinness_ten_nineteen_years    1.701487
Thinness_five_nine_years       1.769183
Schooling                     -0.159410
Life_expectancy               -0.770242
dtype: float64


# measure the kurtosis of the Data
# чем дальше показатель эксцесса (kurtosis) от нуля, тем эксцесс "тяжелее"
life_dropped.kurt()

Alcohol_consumption            -0.732403
Hepatitis_B                     2.721257
Measles                         0.713809
BMI                            -0.150825
Polio                           2.726907
Diphtheria                      3.122858
Incidents_HIV                  28.688935
GDP_per_capita                  6.450232
Population_mln                 70.472298
Thinness_ten_nineteen_years     3.908966
Thinness_five_nine_years        4.302524
Schooling                      -1.039250
Life_expectancy                -0.216851
dtype: float64


# removing outliers
life_less_outliers = life_dropped.copy()
col_names = life_less_outliers.columns.values.tolist()
for each in col_names:
    life_less_outliers.loc[np.abs(stats.zscore(life_less_outliers[each])) > 3, each] = np.NaN
life_less_outliers


# pairplot display
sns.pairplot(life_less_outliers, hue='Life_expectancy', height=2.5,
            kind='scatter')

<seaborn.axisgrid.PairGrid at 0x13d184890>


life.shape

(2864, 21)


# create new dataframe that contains only the numerical columns
life_num = life.select_dtypes(include=np.number)
life_num

# alternately we could convert datatypes to float
#life = life.astype(float)


# combine columns with the Developed and Developing countries status to one new column
life_devstatus = life_num.copy()

life_devstatus['Economy_status_Developed'].replace(0, '', inplace=True)
life_devstatus['Economy_status_Developing'].replace(1, '0', inplace=True)

life_devstatus["Economy_status"] = life_devstatus['Economy_status_Developed'] + life_devstatus['Economy_status_Developing']
life_devstatus = life_devstatus.drop(['Economy_status_Developed','Economy_status_Developing'], axis=1)
life_devstatus


print('Developing countries with a status of 0 are depicted with blue color, \nwhile Developed countries with a status of 1 are represented by orange color.')
sns.lineplot(data=life_devstatus, x='Year', y='Life_expectancy', hue='Economy_status')
plt.legend(title='Economy_status',loc='center')
plt.show()

Developing countries with a status of 0 are depicted with blue color, 
while Developed countries with a status of 1 are represented by orange color.


# split set to Developing and Developed sets
life_developed = life_num[life['Economy_status_Developed'] == 1]
life_developing = life_num[life['Economy_status_Developing'] == 1]

# drop unnecessary columns from new datasets
life_developed = life_developed.drop(['Economy_status_Developed','Economy_status_Developing'],axis=1)
life_developing = life_developing.drop(['Economy_status_Developed','Economy_status_Developing'],axis=1)

print('Developed countries:', life_developed.shape)
print('Developing countries:', life_developing.shape)

Developed countries: (592, 17)
Developing countries: (2272, 17)


# drop unnecessary columns
life_developed = life_developed.drop(['Year','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln'],axis=1)
display(life_developed)


# correlation matrix and heatmap for developed countries
plt.figure(figsize=(15,10))
developed_corrm = life_developed.corr().round(3)
print('With this visualization, we can gain insights into the correlations among various factors.')
sns.heatmap(developed_corrm,annot=True,cmap='Greens')
plt.show()

With this visualization, we can gain insights into the correlations among various factors.


# collect developed column names to list
col_developed = developed_corrm.columns.values.tolist()
col_developed

['Alcohol_consumption',
 'Hepatitis_B',
 'Measles',
 'BMI',
 'Polio',
 'Diphtheria',
 'Incidents_HIV',
 'GDP_per_capita',
 'Thinness_ten_nineteen_years',
 'Thinness_five_nine_years',
 'Schooling',
 'Life_expectancy']


# removing values with no strong correlation
# 0.5+ - weak correlation
# 0.7+ - some correlation
# 0.9+ - strong correlation

for each in col_developed:
    developed_corrm.loc[(developed_corrm[each] > -0.5) & (developed_corrm[each] < 0.5), each] = ''
    developed_corrm.loc[(developed_corrm[each] == 1), each] = ''
developed_corrm.round(3)


# combine thinnes factor columns to one new Thinness under twenty column
# dropping unnecessary columns with no correllation
life_developed['Thinness_under_twenty'] = life_developed[['Thinness_ten_nineteen_years','Thinness_five_nine_years']].mean(axis=1)
life_developed = life_developed.drop(['Thinness_ten_nineteen_years','Thinness_five_nine_years'], axis=1)
developed_short = life_developed[['Life_expectancy','GDP_per_capita','Thinness_under_twenty']]


# create most significant metrics
developed_short_corrm = developed_short.corr().round(3)
developed_short_corrm = developed_short_corrm.drop(['Life_expectancy'], axis=1)
developed_short_corrm = developed_short_corrm.drop(['GDP_per_capita','Thinness_under_twenty'], axis=0)
print('Most significant metrics for developed countries:')
display(developed_short_corrm)

Most significant metrics for developed countries:


# visualize regression
print('Regression for developed countries:')
sns.pairplot(developed_short, x_vars=['GDP_per_capita','Thinness_under_twenty'], y_vars = 'Life_expectancy',
             height=5, kind='reg', plot_kws={'line_kws':{'color':'red'}})
plt.show()

Regression for developed countries:


# combine all immunisation columns to one new Immunisation column
# drop unneccesary colums
life_developing['Immunisation'] = life_developing[['Hepatitis_B','Measles','Polio','Diphtheria']].mean(axis=1)
life_developing = life_developing.drop(['Year','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Hepatitis_B','Measles','Polio','Diphtheria'],axis=1)
display(life_developing)


# correlation matrix and heatmap for developing countries
plt.figure(figsize=(15,10))
developing_corrm = life_developing.corr().round(3)
print('With this visualization, we can gain insights into the correlations among various factors.')
sns.heatmap(developing_corrm,annot=True,cmap='Reds')
plt.show()

With this visualization, we can gain insights into the correlations among various factors.


# collect developing column names to list
col_developing = developing_corrm.columns.values.tolist()
col_developing

['Alcohol_consumption',
 'BMI',
 'Incidents_HIV',
 'GDP_per_capita',
 'Thinness_ten_nineteen_years',
 'Thinness_five_nine_years',
 'Schooling',
 'Life_expectancy',
 'Immunisation']


# removing values with no strong correlation
# 0.5+ - weak correlation
# 0.7+ - some correlation
# 0.9+ - strong correlation

for each in col_developing:
    developing_corrm.loc[(developing_corrm[each] >= -0.5) & (developing_corrm[each] <= 0.5), each] = ''
    developing_corrm.loc[(developing_corrm[each] == 1), each] = ''
developing_corrm.round(3)


# create new short df with most correlating factors
# create new correlation matrix
developing_short = life_developing[['BMI','Incidents_HIV','Schooling','Immunisation','Life_expectancy']]
developing_short_corrm = developing_short.corr().round(3)


# create most significant metrics
developing_short_corrm = developing_short_corrm.drop(['BMI','Incidents_HIV','Schooling','Immunisation'], axis=0)
developing_short_corrm = developing_short_corrm.drop(['Life_expectancy'], axis=1)

print('Most significant metrics for developing countries:')
display(developing_short_corrm)

Most significant metrics for developing countries:


# visualize regression
print('Regression for developing countries:')
sns.pairplot(developing_short, x_vars=['BMI','Incidents_HIV','Schooling','Immunisation'], y_vars = 'Life_expectancy',
             height=5, kind='reg', plot_kws={'line_kws':{'color':'red'}})
plt.show()

Regression for developing countries:


# split the Data to the independent variables and the dependant variable
X_developed = life_developed[['Alcohol_consumption','BMI','Incidents_HIV','GDP_per_capita','Thinness_under_twenty','Schooling']]
y_developed = life_developed.loc[:, 'Life_expectancy']


# split the Data to train and test sets
X_dped_train, X_dped_test, y_dped_train, y_dped_test = train_test_split(X_developed,y_developed, test_size=0.25, random_state=44)    
print('Train data shape:',X_dped_train.shape)
print('Test data shape:',X_dped_test.shape)

Train data shape: (444, 6)
Test data shape: (148, 6)


# fit the training set into the linear regression model
reg_dped = linear_model.LinearRegression()
reg_dped.fit(X_dped_train,y_dped_train)

print("Coefficients:" ,reg_dped.coef_)
print("Intercept:", reg_dped.intercept_)

Coefficients: [-2.63258561e-01 -8.11202340e-01 -9.70034231e-01  1.95627310e-05
 -2.80806713e+00  3.26479838e-01]
Intercept: 101.47680560560815


#calculate predictions
y_dped_pred= pd.DataFrame(reg_dped.predict(X_dped_test))
y_dped_pred['index'] = X_dped_test.index
y_dped_pred.set_index('index', inplace=True, drop=True)
print(y_dped_pred)

               0
index           
1056   74.276877
1034   75.238431
166    75.185610
899    80.976138
2569   79.563017
...          ...
1983   80.759134
1336   77.908321
2768   80.796583
2249   80.550230
767    80.131220

[148 rows x 1 columns]


# count the accuracy of the model with the test data
reg_mse_dped = mean_squared_error(y_dped_test, y_dped_pred)
reg_r2_dped = r2_score(y_dped_test, y_dped_pred)


# create and fit the Lasso Regression model
lasso_dped = Lasso(alpha=0.001)  # medium alpha
lasso_dped.fit(X_dped_train, y_dped_train)

# make predictions
y_dped_pred = lasso_dped.predict(X_dped_test)

# evaluate the model
lasso_mse_dped = mean_squared_error(y_dped_test, y_dped_pred)
lasso_r2_dped = r2_score(y_dped_test, y_dped_pred)


# fit the training set into the ridge regression model
ridge = Ridge()
kf = KFold()
cv_ridge = cross_val_score(ridge, X_developed, y_developed, cv=kf)
ridge_dped = Ridge(alpha=0.001)
ridge_dped.fit(X_dped_train, y_dped_train)
y_dped_pred_ridge = ridge_dped.predict(X_dped_test)

ridge_mse_dped = mean_squared_error(y_dped_test, y_dped_pred)
ridge_r2_dped = r2_score(y_dped_test, y_dped_pred)


print('Mean Squared Error and R-squared results')
print('Lower MSE values indicate better model performance')
print('Higher R2 values indicate a better fit of the model to the data')
print('')
print('Linear Regression Test dataset result:')
print("MSE = ", reg_mse_dped)
print("R2s = ", reg_r2_dped)
print('')
print('Lasso Regression Test dataset result:')
print("MSE:", lasso_mse_dped)
print("R2:", lasso_r2_dped)
print('')
print('Ridge Regression Test dataset result:')
print("MSE:", ridge_mse_dped)
print("R2:", ridge_r2_dped)

Mean Squared Error and R-squared results
Lower MSE values indicate better model performance
Higher R2 values indicate a better fit of the model to the data

Linear Regression Test dataset result:
MSE =  2.663625775877976
R2s =  0.7525790000288742

Lasso Regression Test dataset result:
MSE: 2.6546823451247246
R2: 0.7534097445727141

Ridge Regression Test dataset result:
MSE: 2.6546823451247246
R2: 0.7534097445727141


from sklearn.linear_model import LassoCV
from sklearn.model_selection import cross_val_score

# Create LassoCV model
reg_lasso = LassoCV(alphas=[0.001, 0.01, 0.1, 1.0, 10.0])  # the alpha values could be adjust as needed

# Linear Regression cross-validation
las_k3 = 3
las_scores_dped_3 = cross_val_score(estimator=reg_lasso,
                               X=X_dped_test,
                               y=y_dped_test,
                               cv=las_k3)
las_k5 = 5
las_scores_dped_5 = cross_val_score(estimator=reg_lasso,
                               X=X_dped_test,
                               y=y_dped_test,
                               cv=las_k5)
las_k10 = 10
las_scores_dped_10 = cross_val_score(estimator=reg_lasso,
                               X=X_dped_test,
                               y=y_dped_test,
                               cv=las_k10)

print('Performance of the Lasso Regression model with 3, 5 and 10 folds.')
print()
print('Performance of the model with', las_k3, 'individual folds')
print(las_scores_dped_3)
print("Model performance based on %d-fold cross validation = %.3f" % (las_k3, las_scores_dped_3.mean()))
print()
print('Performance of the model with', las_k5, 'individual folds')
print(las_scores_dped_5)
print("Model performance based on %d-fold cross validation = %.3f" % (las_k5, las_scores_dped_5.mean()))
print()
print('Performance of the model with', las_k10, 'individual folds')
print(las_scores_dped_10)
print("Model performance based on %d-fold cross validation = %.3f" % (las_k10, las_scores_dped_10.mean()))

Performance of the Lasso Regression model with 3, 5 and 10 folds.

Performance of the model with 3 individual folds
[0.69816594 0.71074497 0.74807   ]
Model performance based on 3-fold cross validation = 0.719

Performance of the model with 5 individual folds
[0.65784913 0.79620236 0.71497912 0.68899166 0.71584609]
Model performance based on 5-fold cross validation = 0.715

Performance of the model with 10 individual folds
[0.77280843 0.67783878 0.79357494 0.8047035  0.80098719 0.61890635
 0.79352721 0.53028228 0.68871556 0.68031433]
Model performance based on 10-fold cross validation = 0.716


# calculate residuals (difference between the observed values and the predicted values)
residuals_dped = pd.DataFrame(data = y_dped_test)['Life_expectancy']-y_dped_pred[0]
print(residuals_dped)

1056   -1.483944
1034    0.916056
166     0.216056
899     7.316056
2569    6.116056
          ...   
1983    6.416056
1336    2.816056
2768    4.816056
2249    4.816056
767     6.516056
Name: Life_expectancy, Length: 148, dtype: float64


# set y_dped_pred as df
y_dped_pred = pd.DataFrame(y_dped_pred)
# reset the index of residuals_dped
residuals_dped.reset_index(drop=True, inplace=True)

# concatenate the DataFrames
residuals_dped_tbl = pd.concat([y_dped_test.reset_index(drop=True), y_dped_pred, residuals_dped], axis=1)
residuals_dped_tbl.columns = ['observed', 'predicted', 'residual']
print(residuals_dped_tbl)

     observed  predicted  residual
0        72.8  74.283944 -1.483944
1        75.2  75.242175  0.916056
2        74.5  75.319110  0.216056
3        81.6  80.992415  7.316056
4        80.4  79.565435  6.116056
..        ...        ...       ...
143      80.7  80.728151  6.416056
144      77.1  77.986097  2.816056
145      79.1  80.770243  4.816056
146      79.1  80.524975  4.816056
147      80.8  80.143815  6.516056

[148 rows x 3 columns]


# check the distribution of the predicted and observed values
plt.figure()
residuals_dped_tbl.plot.scatter(x='observed', y='predicted')
plt.plot(y_dped_pred, y_dped_pred, color='red', linewidth=1)

plt.show()

<Figure size 640x480 with 0 Axes>


# check the distribution of residuals and observed values
residuals_dped_tbl.plot.scatter(x='observed', y='residual')
plt.plot([40,85], [0,0], color='red', linewidth=1)
plt.show()


# split the Data to the independent variables and the dependant variable

X_developing = life_developing.drop(life_developing.columns[7], axis=1)
y_developing = life_developing.iloc[:,7]


# split the Data to train and test sets

X_ding_train, X_ding_test, y_ding_train, y_ding_test = train_test_split(X_developing,y_developing, test_size=0.25, random_state=44)    
print('Train data shape:',X_ding_train.shape)
print('Test data shape:',X_ding_test.shape)

Train data shape: (1704, 8)
Test data shape: (568, 8)


# fit the training set into the linear regression model

reg_ding = linear_model.LinearRegression()
reg_ding.fit(X_ding_train,y_ding_train)

print("Coefficients:" ,reg_ding.coef_)
print("Intercept:", reg_ding.intercept_)

Coefficients: [-5.11070954e-02  8.77566679e-01 -1.52865189e+00  1.14594692e-04
 -5.68897641e-02  2.59164979e-01  8.29776845e-01  1.97409970e-01]
Intercept: 22.990765864435495


#calculate predictions
y_ding_pred= pd.DataFrame(reg_ding.predict(X_ding_test))
y_ding_pred['index'] = X_ding_test.index
y_ding_pred.set_index('index', inplace=True, drop=True)
print(y_ding_pred)

               0
index           
1717   67.082742
2287   73.441167
862    63.693050
425    75.043715
2136   62.873420
...          ...
200    64.588438
1474   59.966735
330    67.846597
1464   60.978702
812    55.923640

[568 rows x 1 columns]


# check the accuracy of the model with the test data

mse_ding = mean_squared_error(y_ding_test, y_ding_pred)
r2score_ding = r2_score(y_ding_test, y_ding_pred)


# create and fit the Lasso Regression model
lasso_ding = Lasso(alpha=0.001)  # medium alpha
lasso_ding.fit(X_ding_train, y_ding_train)

# make predictions
y_ding_pred = lasso_ding.predict(X_ding_test)

# evaluate the model
lasso_mse_ding = mean_squared_error(y_ding_test, y_ding_pred)
lasso_r2_ding = r2_score(y_ding_test, y_ding_pred)


# fit the training set into the ridge regression model

ridge = Ridge()
kf = KFold()
cv_ridge = cross_val_score(ridge, X_developing, y_developing, cv=kf)
ridge_ding = Ridge(alpha=0.001)
ridge_ding.fit(X_ding_train, y_ding_train)
y_ding_pred_ridge = ridge_ding.predict(X_ding_test)

ridge_mse_ding = mean_squared_error(y_ding_test, y_ding_pred)
ridge_r2_ding = r2_score(y_ding_test, y_ding_pred)


print('Mean Squared Error and R-squared results')
print('Lower MSE values indicate better model performance')
print('Higher R2 values indicate a better fit of the model to the data')
print('')
print('Linear Regression Test dataset result:')
print("MSE = ", mse_ding)
print("R2s = ", r2score_ding)
print('')
print('Lasso Regression Test dataset result:')
print("MSE:", lasso_mse_ding)
print("R2:", lasso_r2_ding)
print('')
print('Ridge Regression Test dataset result:')
print("MSE:", ridge_mse_ding)
print("R2:", ridge_r2_ding)

# MSE - меньше (ближе к 0) лучше, R2s - больше (ближе к 1 из 0-1) лучше

Mean Squared Error and R-squared results
Lower MSE values indicate better model performance
Higher R2 values indicate a better fit of the model to the data

Linear Regression Test dataset result:
MSE =  16.208529584359944
R2s =  0.7895569104640856

Lasso Regression Test dataset result:
MSE: 16.209230681162026
R2: 0.789547807789082

Ridge Regression Test dataset result:
MSE: 16.209230681162026
R2: 0.789547807789082


# Linear Regression cross-validation
from sklearn.model_selection import cross_val_score
lin_k3 = 3
lin_scores_ding_3 = cross_val_score(estimator=reg_ding,
                        X=X_ding_test,
                        y=y_ding_test,
                        cv=lin_k3)
lin_k5 = 5
lin_scores_ding_5 = cross_val_score(estimator=reg_ding,
                        X=X_ding_test,
                        y=y_ding_test,
                        cv=lin_k5)
lin_k10 = 10
lin_scores_ding_10 = cross_val_score(estimator=reg_ding,
                        X=X_ding_test,
                        y=y_ding_test,
                        cv=lin_k10)

print('Performance of the Linear Regression model with 3, 5 and 10 folds.')
print()
print('Performance of the model with', lin_k3, 'individual folds')
print(lin_scores_ding_3)
print("Model performance based on %d-fold cross validation = %.3f" % (lin_k3, lin_scores_ding_3.mean()))
print()
print('Performance of the model with', lin_k5, 'individual folds')
print(lin_scores_ding_5)
print("Model performance based on %d-fold cross validation = %.3f" % (lin_k5, lin_scores_ding_5.mean()))
print()
print('Performance of the model with', lin_k10, 'individual folds')
print(lin_scores_ding_10)
print("Model performance based on %d-fold cross validation = %.3f" % (lin_k10, lin_scores_ding_10.mean()))

Performance of the Linear Regression model with 3, 5 and 10 folds.

Performance of the model with 3 individual folds
[0.78507755 0.76690012 0.76636384]
Model performance based on 3-fold cross validation = 0.773

Performance of the model with 5 individual folds
[0.79047868 0.79096892 0.78367983 0.69784204 0.80794088]
Model performance based on 5-fold cross validation = 0.774

Performance of the model with 10 individual folds
[0.75005117 0.82112522 0.77467799 0.81261227 0.82498273 0.7531495
 0.68660956 0.73724225 0.77532502 0.84052676]
Model performance based on 10-fold cross validation = 0.778


# calculate residuals (difference between the observed values and the predicted values)
residuals_ding = pd.DataFrame(data = y_ding_test)['Life_expectancy']-y_ding_pred[0]
print(residuals_ding)

1717     4.517671
2287     0.617671
862     -1.682329
425      7.917671
2136    -2.282329
          ...    
200      1.517671
1474    -6.582329
330      2.817671
1464   -10.282329
812    -12.182329
Name: Life_expectancy, Length: 568, dtype: float64


# set y_dped_pred as df
y_ding_pred = pd.DataFrame(y_ding_pred)
# reset the index of residuals_dped
residuals_ding.reset_index(drop=True, inplace=True)

# concatenate the DataFrames
residuals_ding_tbl = pd.concat([y_ding_test.reset_index(drop=True), y_ding_pred, residuals_ding], axis=1)
residuals_ding_tbl.columns = ['observed', 'predicted', 'residual']
print(residuals_ding_tbl)

     observed  predicted   residual
0        71.6  67.082329   4.517671
1        67.7  73.441004   0.617671
2        65.4  63.694569  -1.682329
3        75.0  75.043717   7.917671
4        64.8  62.874090  -2.282329
..        ...        ...        ...
563      68.6  64.588899   1.517671
564      60.5  59.966796  -6.582329
565      69.9  67.845072   2.817671
566      56.8  60.978901 -10.282329
567      54.9  55.924585 -12.182329

[568 rows x 3 columns]


# check the distribution of the predicted and observed values
plt.figure()
residuals_ding_tbl.plot.scatter(x='observed', y='predicted')
plt.plot(y_ding_pred, y_ding_pred, color='red', linewidth=1)

plt.show()

<Figure size 640x480 with 0 Axes>


# check the distribution of residuals and observed values
residuals_ding_tbl.plot.scatter(x='observed', y='residual')
plt.scatter(x=pd.DataFrame(data = y_ding_test)['Life_expectancy'], y=residuals_ding)
plt.plot([40,85], [0,0], color='red', linewidth=1)
plt.show()


# create dataframes based on Regions
print(life['Region'].unique())

life_Africa = life[life['Region'] == 'Africa']
life_Africa = life_Africa.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Middle_East = life[life['Region'] == 'Middle East']
life_Middle_East = life_Middle_East.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Europe = life[life['Region'] == 'European Union']
life_Europe = life_Europe.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Asia = life[life['Region'] == 'Asia']
life_Asia = life_Asia.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Oceania = life[life['Region'] == 'Oceania']
life_Oceania = life_Oceania.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Rest_of_Europe = life[life['Region'] == 'Rest of Europe']
life_Rest_of_Europe = life_Rest_of_Europe.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_Central_America_and_Caribbean = life[life['Region'] == 'Central America and Caribbean']
life_Central_America_and_Caribbean = life_Central_America_and_Caribbean.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_North_America = life[life['Region'] == 'North America']
life_North_America = life_North_America.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)
life_South_America = life[life['Region'] == 'South America']
life_South_America = life_South_America.drop(['Country','Infant_deaths','Under_five_deaths','Adult_mortality','Population_mln','Economy_status_Developed','Economy_status_Developing'], axis=1)

['Middle East' 'European Union' 'Asia' 'South America'
 'Central America and Caribbean' 'Rest of Europe' 'Africa' 'Oceania'
 'North America']


# concat region datasets to new df
lifes = [life_Africa, life_Middle_East, life_Europe, life_Asia, life_Oceania,
       life_Rest_of_Europe, life_Central_America_and_Caribbean,
       life_North_America, life_South_America]

life_regions = pd.concat(lifes, ignore_index=True)
life_regions


# group by 'Region' and 'Year' and calculate the mean for each one
life_regions_mean = life_regions.groupby(['Region','Year']).mean().reset_index()

# round the values to two decimal places
life_regions_mean = life_regions_mean.round(2)
life_regions_mean


import plotly.express as px
# to make scatters visible in html format
from plotly.offline import init_notebook_mode
init_notebook_mode(connected=True)


# pip install --upgrade nbformat


px.scatter(life_regions_mean,
           y= "Alcohol_consumption",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Alcohol_consumption',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Alcohol_consumption'
           )


px.scatter(life_regions_mean,
           y= "Incidents_HIV",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Incidents_HIV',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Incidents_HIV'
            )


px.scatter(life_regions_mean,
           y= "GDP_per_capita",
           x= 'Life_expectancy',
           animation_group='Region',
           size='GDP_per_capita',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to GDP_per_capita'
            )


px.scatter(life_regions_mean,
           y= "Hepatitis_B",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Hepatitis_B',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Hepatitis_B'
            )


px.scatter(life_regions_mean,
           y= "BMI",
           x= 'Life_expectancy',
           animation_group='Region',
           size='BMI',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to BMI'
            )


px.scatter(life_regions_mean,
           y= "Measles",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Measles',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Measles'
            )


px.scatter(life_regions_mean,
           y= "Diphtheria",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Diphtheria',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Diphtheria'
            )


px.scatter(life_regions_mean,
           y= "Polio",
           x= 'Life_expectancy',
           animation_group='Region',
           size='Polio',
           color='Region',
           template='plotly_dark',
           title='<b> Life Expectancy in Relation to Polio'
            )


px.scatter(life_regions_mean.sort_values(by='Year'),
           y='Schooling',
           x='Life_expectancy',
           animation_frame='Year',
           animation_group='Region',
           template='plotly_dark',
           color='Region',
           size='Life_expectancy',
           title='<b>Life Expectancy in Relation to Schooling in Every Year'
           )


# make a table for the VIF values of all region based data frames' X variables.
from statsmodels.stats.outliers_influence import variance_inflation_factor


life_Africa_num = life_Africa.select_dtypes(include=np.number)

X_Africa = life_Africa_num.drop(['Life_expectancy'], axis = 1).values
y_Africa = life_Africa_num['Life_expectancy'].values
names_Africa = life_Africa_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Africa = pd.DataFrame()
vif_Africa["VIF Factor"] = [variance_inflation_factor(X_Africa, i) for i in range(names_Africa.shape[0])]
vif_Africa["features"] = names_Africa
vif_Africa.round()


life_Middle_East_num = life_Middle_East.select_dtypes(include=np.number)

X_Middle_East = life_Middle_East_num.drop(['Life_expectancy'], axis = 1).values
y_Middle_East = life_Middle_East_num['Life_expectancy'].values
names_Middle_East = life_Middle_East_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Middle_East = pd.DataFrame()
vif_Middle_East["VIF Factor"] = [variance_inflation_factor(X_Middle_East, i) for i in range(names_Middle_East.shape[0])]
vif_Middle_East["features"] = names_Middle_East
vif_Middle_East.round()


life_Europe_num = life_Europe.select_dtypes(include=np.number)

X_Europe = life_Europe_num.drop(['Life_expectancy'], axis = 1).values
y_Europe = life_Europe_num['Life_expectancy'].values
names_Europe = life_Europe_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Europe = pd.DataFrame()
vif_Europe["VIF Factor"] = [variance_inflation_factor(X_Europe, i) for i in range(names_Europe.shape[0])]
vif_Europe["features"] = names_Europe
vif_Europe.round()


life_Asia_num = life_Asia.select_dtypes(include=np.number)

X_Asia = life_Asia_num.drop(['Life_expectancy'], axis = 1).values
y_Asia = life_Asia_num['Life_expectancy'].values
names_Asia = life_Asia_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Asia = pd.DataFrame()
vif_Asia["VIF Factor"] = [variance_inflation_factor(X_Asia, i) for i in range(names_Asia.shape[0])]
vif_Asia["features"] = names_Asia
vif_Asia.round()


life_Oceania_num = life_Oceania.select_dtypes(include=np.number)

X_Oceania = life_Oceania_num.drop(['Life_expectancy'], axis = 1).values
y_Oceania = life_Oceania_num['Life_expectancy'].values
names_Oceania = life_Oceania_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Oceania = pd.DataFrame()
vif_Oceania["VIF Factor"] = [variance_inflation_factor(X_Oceania, i) for i in range(names_Oceania.shape[0])]
vif_Oceania["features"] = names_Oceania
vif_Oceania.round()


life_Rest_of_Europe_num = life_Rest_of_Europe.select_dtypes(include=np.number)

X_Rest_of_Europe = life_Rest_of_Europe_num.drop(['Life_expectancy'], axis = 1).values
y_Rest_of_Europe = life_Rest_of_Europe_num['Life_expectancy'].values
names_Rest_of_Europe = life_Rest_of_Europe_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Rest_of_Europe = pd.DataFrame()
vif_Rest_of_Europe["VIF Factor"] = [variance_inflation_factor(X_Rest_of_Europe, i) for i in range(names_Rest_of_Europe.shape[0])]
vif_Rest_of_Europe["features"] = names_Rest_of_Europe
vif_Rest_of_Europe.round()


life_Central_America_and_Caribbean_num = life_Central_America_and_Caribbean.select_dtypes(include=np.number)

X_Central_America_and_Caribbean = life_Central_America_and_Caribbean_num.drop(['Life_expectancy'], axis = 1).values
y_Central_America_and_Caribbean = life_Central_America_and_Caribbean_num['Life_expectancy'].values
names_Central_America_and_Caribbean = life_Central_America_and_Caribbean_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_Central_America_and_Caribbean = pd.DataFrame()
vif_Central_America_and_Caribbean["VIF Factor"] = [variance_inflation_factor(X_Central_America_and_Caribbean, i) for i in range(names_Central_America_and_Caribbean.shape[0])]
vif_Central_America_and_Caribbean["features"] = names_Central_America_and_Caribbean
vif_Central_America_and_Caribbean.round()


life_North_America_num = life_North_America.select_dtypes(include=np.number)

X_North_America = life_North_America_num.drop(['Life_expectancy'], axis = 1).values
y_North_America = life_North_America_num['Life_expectancy'].values
names_North_America = life_North_America_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_North_America = pd.DataFrame()
vif_North_America["VIF Factor"] = [variance_inflation_factor(X_North_America, i) for i in range(names_North_America.shape[0])]
vif_North_America["features"] = names_North_America
vif_North_America.round()


life_South_America_num = life_South_America.select_dtypes(include=np.number)

X_South_America = life_South_America_num.drop(['Life_expectancy'], axis = 1).values
y_South_America = life_South_America_num['Life_expectancy'].values
names_South_America = life_South_America_num.drop(['Year','Life_expectancy'], axis = 1).columns

vif_South_America = pd.DataFrame()
vif_South_America["VIF Factor"] = [variance_inflation_factor(X_South_America, i) for i in range(names_South_America.shape[0])]
vif_South_America["features"] = names_South_America
vif_South_America.round()

	count	mean	std	min	25%	50%	75%	max
Year	2864.0	2007.500000	4.610577	2000.000	2003.75000	2007.5000	2011.250000	2015.0000
Infant_deaths	2864.0	30.363792	27.538117	1.800	8.10000	19.6000	47.350000	138.1000
Under_five_deaths	2864.0	42.938268	44.569974	2.300	9.67500	23.1000	66.000000	224.9000
Adult_mortality	2864.0	192.251775	114.910281	49.384	106.91025	163.8415	246.791375	719.3605
Alcohol_consumption	2864.0	4.820882	3.981949	0.000	1.20000	4.0200	7.777500	17.8700
Hepatitis_B	2864.0	84.292598	15.995511	12.000	78.00000	89.0000	96.000000	99.0000
Measles	2864.0	77.344972	18.659693	10.000	64.00000	83.0000	93.000000	99.0000
BMI	2864.0	25.032926	2.193905	19.800	23.20000	25.5000	26.400000	32.1000
Polio	2864.0	86.499651	15.080365	8.000	81.00000	93.0000	97.000000	99.0000
Diphtheria	2864.0	86.271648	15.534225	16.000	81.00000	93.0000	97.000000	99.0000
Incidents_HIV	2864.0	0.894288	2.381389	0.010	0.08000	0.1500	0.460000	21.6800
GDP_per_capita	2864.0	11540.924930	16934.788931	148.000	1415.75000	4217.0000	12557.000000	112418.0000
Population_mln	2864.0	36.675915	136.485867	0.080	2.09750	7.8500	23.687500	1379.8600
Thinness_ten_nineteen_years	2864.0	4.865852	4.438234	0.100	1.60000	3.3000	7.200000	27.7000
Thinness_five_nine_years	2864.0	4.899825	4.525217	0.100	1.60000	3.4000	7.300000	28.6000
Schooling	2864.0	7.632123	3.171556	1.100	5.10000	7.8000	10.300000	14.1000
Economy_status_Developed	2864.0	0.206704	0.405012	0.000	0.00000	0.0000	0.000000	1.0000
Economy_status_Developing	2864.0	0.793296	0.405012	0.000	1.00000	1.0000	1.000000	1.0000
Life_expectancy	2864.0	68.856075	9.405608	39.400	62.70000	71.4000	75.400000	83.8000

	Alcohol_consumption	Hepatitis_B	Measles	BMI	Polio	Diphtheria	Incidents_HIV	GDP_per_capita	Population_mln	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy
0	1.320	97.0	65.0	27.8	97.0	97.0	0.08	11006.0	78.53	4.9	4.8	7.8	76.5
1	10.350	97.0	94.0	26.0	97.0	97.0	0.09	25742.0	46.44	0.6	0.5	9.7	82.8
2	1.570	60.0	35.0	21.2	67.0	64.0	0.13	1076.0	NaN	NaN	NaN	5.0	65.4
3	5.680	93.0	74.0	25.3	92.0	93.0	0.79	4146.0	0.75	5.7	5.5	7.9	67.0
4	2.890	97.0	89.0	27.0	94.0	94.0	0.08	33995.0	7.91	1.2	1.1	12.8	81.7
...	...	...	...	...	...	...	...	...	...	...	...	...	...
2859	0.092	72.0	64.0	20.8	NaN	NaN	0.49	399.0	11.33	12.8	12.9	1.1	49.9
2860	6.560	97.0	97.0	25.3	96.0	95.0	0.02	2515.0	2.67	2.2	2.3	9.1	66.9
2861	1.560	62.0	95.0	21.9	97.0	97.0	0.02	2130.0	19.39	15.4	15.5	10.3	74.3
2862	11.000	94.0	95.0	26.1	97.0	95.0	0.05	7424.0	3.44	3.3	3.3	11.1	71.8
2863	6.840	88.0	90.0	26.1	95.0	95.0	0.05	48934.0	0.32	0.9	0.9	11.0	82.4

	Alcohol_consumption	BMI	Incidents_HIV	GDP_per_capita	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy	Immunisation
0	1.320	27.8	0.08	11006	4.9	4.8	7.8	76.5	89.00
2	1.570	21.2	0.13	1076	27.1	28.0	5.0	65.4	56.50
3	5.680	25.3	0.79	4146	5.7	5.5	7.9	67.0	88.00
5	4.190	26.4	0.16	9110	2.0	1.9	7.9	78.2	88.00
6	8.060	26.2	0.08	9313	2.3	2.3	12.0	71.2	97.00
...	...	...	...	...	...	...	...	...	...
2857	12.920	25.7	0.05	2617	2.7	2.8	8.9	68.9	90.00
2858	0.023	25.6	0.14	5305	7.5	7.8	8.9	72.6	89.75
2859	0.092	20.8	0.49	399	12.8	12.9	1.1	49.9	52.75
2860	6.560	25.3	0.02	2515	2.2	2.3	9.1	66.9	96.25
2861	1.560	21.9	0.02	2130	15.4	15.5	10.3	74.3	87.75

	Region	Year	Alcohol_consumption	Hepatitis_B	Measles	BMI	Polio	Diphtheria	Incidents_HIV	GDP_per_capita	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy
0	Africa	2000	3.06	72.96	67.06	22.65	65.43	64.69	3.86	1890.78	7.90	7.74	3.95	54.14
1	Africa	2001	3.06	73.90	66.65	22.75	66.16	65.67	3.65	1942.47	7.35	7.18	4.01	54.36
2	Africa	2002	3.03	72.78	67.08	22.83	69.84	68.10	3.46	1971.27	7.86	7.52	4.11	54.59
3	Africa	2003	2.91	74.39	67.31	22.91	71.75	70.57	3.29	2016.45	7.32	7.32	4.17	54.96
4	Africa	2004	2.86	75.31	67.49	23.00	72.63	71.88	3.14	2113.61	7.48	7.64	4.26	55.45
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
139	South America	2011	5.75	90.25	76.17	26.41	89.83	90.33	0.27	8257.08	2.07	1.97	8.44	73.70
140	South America	2012	5.92	90.75	74.83	26.50	89.33	90.50	0.27	8587.92	2.04	1.96	8.48	73.92
141	South America	2013	5.78	90.50	72.50	26.59	88.42	90.67	0.26	8778.17	2.03	1.93	8.62	74.11
142	South America	2014	5.74	90.58	75.42	26.68	89.50	90.33	0.26	9136.67	2.02	1.93	8.76	74.27
143	South America	2015	5.61	92.00	77.33	26.79	91.17	91.92	0.25	9229.50	2.02	1.92	8.84	74.47

EDA (Exploratory Data Analysis) with linear regression by Natalia Lappalainen - 10/2023¶

The aim of this analysis is to determine the factors that influence life expectancy.¶

Used source:¶

Key factors:¶

Hypotheses:¶

1. Extract and explore dataset¶

2. Transforming the Data¶

Skewness and kurtosis¶

Next, I decided to divide and explore unconverted Data as follows:¶

1) Status-based analysis (Developed, Developing)¶

Let's check how Life expectancy differ between Developed and Developing countries in general.¶

We can observe a substantial disparity in life expectancy between developed and developing countries.¶

Let's explore Developed and Developing countries separately.¶

1.1 Developed countries¶

1.2 Developing countries¶

3. Modeling¶

Now, we need to split our DataFrames into the independent (predictor) variables X and the continuous dependent (response) variable Y¶

Developed countries¶

Performance of the Lasso Regression model with 3 individual folds shows us less variability across different folds and the highest average performance. Thus model with 3 folds could be considered as the best option in this case.¶

Developing countries¶

Performance of the Linear Regression model with 3 individual folds shows us less variability across different folds. The model with 10 folds shows the highest average performance.¶

2) Region-based analysis (Africa, Middle East, European Union, Rest of Europe, Asia, Oceania, Central America and Caribbean, North America, South America)¶

Presenting few scatters to visualize data¶

Next we can check VIF (Variance Inflation Factor)¶

Final Conclusion:¶

Developed Countries:¶

Developing Countries:¶

Regional Variances:¶

Comparison:¶

Implications and Relevance:¶

Future Directions:¶

	Country	Region	Year	Infant_deaths	Under_five_deaths	Adult_mortality	Alcohol_consumption	Hepatitis_B	Measles	BMI	...	Diphtheria	Incidents_HIV	GDP_per_capita	Population_mln	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Economy_status_Developed	Economy_status_Developing	Life_expectancy
0	Turkiye	Middle East	2015	11.1	13.0	105.8240	1.320	97	65	27.8	...	97	0.08	11006	78.53	4.9	4.8	7.8	0	1	76.5
1	Spain	European Union	2015	2.7	3.3	57.9025	10.350	97	94	26.0	...	97	0.09	25742	46.44	0.6	0.5	9.7	1	0	82.8
2	India	Asia	2007	51.5	67.9	201.0765	1.570	60	35	21.2	...	64	0.13	1076	1183.21	27.1	28.0	5.0	0	1	65.4
3	Guyana	South America	2006	32.8	40.5	222.1965	5.680	93	74	25.3	...	93	0.79	4146	0.75	5.7	5.5	7.9	0	1	67.0
4	Israel	Middle East	2012	3.4	4.3	57.9510	2.890	97	89	27.0	...	94	0.08	33995	7.91	1.2	1.1	12.8	1	0	81.7
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2859	Niger	Africa	2000	97.0	224.9	291.8240	0.092	72	64	20.8	...	34	0.49	399	11.33	12.8	12.9	1.1	0	1	49.9
2860	Mongolia	Asia	2009	23.9	28.6	235.2330	6.560	97	97	25.3	...	95	0.02	2515	2.67	2.2	2.3	9.1	0	1	66.9
2861	Sri Lanka	Asia	2004	17.7	28.9	134.8950	1.560	62	95	21.9	...	97	0.02	2130	19.39	15.4	15.5	10.3	0	1	74.3
2862	Lithuania	European Union	2002	7.9	9.9	204.0120	11.000	94	95	26.1	...	95	0.05	7424	3.44	3.3	3.3	11.1	1	0	71.8
2863	Iceland	Rest of Europe	2011	2.1	2.6	50.5745	6.840	88	90	26.1	...	95	0.05	48934	0.32	0.9	0.9	11.0	1	0	82.4

	Alcohol_consumption	Hepatitis_B	Measles	BMI	Polio	Diphtheria	Incidents_HIV	GDP_per_capita	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy
1	10.35	97	94	26.0	97	97	0.09	25742	0.6	0.5	9.7	82.8
4	2.89	97	89	27.0	94	94	0.08	33995	1.2	1.1	12.8	81.7
7	12.23	88	99	25.9	99	99	0.08	8971	2.3	2.3	10.2	71.2
11	7.47	82	77	27.1	76	74	0.08	19338	0.7	0.7	9.9	79.8
14	10.24	88	71	25.0	94	94	0.08	79368	0.6	0.4	12.3	81.5
...	...	...	...	...	...	...	...	...	...	...	...	...
2850	10.10	97	83	25.9	98	98	0.08	39025	0.9	0.9	11.1	80.0
2855	8.94	88	92	25.8	95	98	0.08	38785	0.9	0.8	9.4	78.0
2856	8.40	83	94	22.7	99	96	0.17	34387	2.0	1.7	12.5	83.6
2862	11.00	94	95	26.1	97	95	0.05	7424	3.3	3.3	11.1	71.8
2863	6.84	88	90	26.1	95	95	0.05	48934	0.9	0.9	11.0	82.4

	BMI	Incidents_HIV	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy	Immunisation
Alcohol_consumption
BMI			-0.571	-0.574	0.652	0.593
Incidents_HIV						-0.559
GDP_per_capita
Thinness_ten_nineteen_years	-0.571			0.926
Thinness_five_nine_years	-0.574		0.926
Schooling	0.652					0.636	0.535
Life_expectancy	0.593	-0.559			0.636		0.626
Immunisation					0.535	0.626

	Region	Year	Alcohol_consumption	Hepatitis_B	Measles	BMI	Polio	Diphtheria	Incidents_HIV	GDP_per_capita	Thinness_ten_nineteen_years	Thinness_five_nine_years	Schooling	Life_expectancy
0	Africa	2007	0.15	75	64	23.5	75	75	0.02	1166	7.3	7.2	3.5	60.7
1	Africa	2012	7.47	82	64	24.9	80	82	1.57	7181	6.3	6.2	7.8	62.9
2	Africa	2011	1.67	91	58	23.7	91	91	0.93	1580	6.9	6.8	6.8	61.4
3	Africa	2003	0.90	71	64	22.3	50	50	2.23	2072	9.1	8.8	5.7	53.8
4	Africa	2011	0.89	73	64	21.1	73	73	0.15	464	7.5	7.4	6.1	63.8
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
2859	South America	2002	4.32	78	93	25.0	83	81	0.23	4003	2.5	2.3	6.5	73.5
2860	South America	2015	6.21	87	83	26.7	87	87	0.40	17318	1.6	1.5	10.1	72.6
2861	South America	2010	5.56	95	82	25.8	95	95	0.65	4847	5.6	5.3	8.1	68.1
2862	South America	2012	8.35	90	74	27.6	90	90	0.19	12973	0.8	0.8	9.9	79.2
2863	South America	2000	7.68	73	56	26.1	88	83	0.16	10731	1.2	1.1	9.1	73.6

	VIF Factor	features
0	362.0	Alcohol_consumption
1	3.0	Hepatitis_B
2	95.0	Measles
3	25.0	BMI
4	415.0	Polio
5	197.0	Diphtheria
6	252.0	Incidents_HIV
7	2.0	GDP_per_capita
8	4.0	Thinness_ten_nineteen_years
9	30.0	Thinness_five_nine_years
10	29.0	Schooling

	VIF Factor	features
0	2225.0	Alcohol_consumption
1	7.0	Hepatitis_B
2	211.0	Measles
3	47.0	BMI
4	1784.0	Polio
5	1502.0	Diphtheria
6	1349.0	Incidents_HIV
7	13.0	GDP_per_capita
8	6.0	Thinness_ten_nineteen_years
9	4664.0	Thinness_five_nine_years
10	4536.0	Schooling

	VIF Factor	features
0	3537.0	Alcohol_consumption
1	29.0	Hepatitis_B
2	84.0	Measles
3	145.0	BMI
4	2954.0	Polio
5	6542.0	Diphtheria
6	6708.0	Incidents_HIV
7	4.0	GDP_per_capita
8	6.0	Thinness_ten_nineteen_years
9	386.0	Thinness_five_nine_years
10	328.0	Schooling

	VIF Factor	features
0	389.0	Alcohol_consumption
1	3.0	Hepatitis_B
2	67.0	Measles
3	26.0	BMI
4	402.0	Polio
5	460.0	Diphtheria
6	511.0	Incidents_HIV
7	3.0	GDP_per_capita
8	2.0	Thinness_ten_nineteen_years
9	14.0	Thinness_five_nine_years
10	15.0	Schooling

	VIF Factor	features
0	837.0	Alcohol_consumption
1	32.0	Hepatitis_B
2	290.0	Measles
3	35.0	BMI
4	920.0	Polio
5	204.0	Diphtheria
6	446.0	Incidents_HIV
7	9.0	GDP_per_capita
8	25.0	Thinness_ten_nineteen_years
9	16.0	Thinness_five_nine_years
10	20.0	Schooling

	VIF Factor	features
0	4817.0	Alcohol_consumption
1	11.0	Hepatitis_B
2	101.0	Measles
3	179.0	BMI
4	4944.0	Polio
5	1571.0	Diphtheria
6	1120.0	Incidents_HIV
7	3.0	GDP_per_capita
8	10.0	Thinness_ten_nineteen_years
9	829.0	Thinness_five_nine_years
10	730.0	Schooling

	VIF Factor	features
0	1210.0	Alcohol_consumption
1	16.0	Hepatitis_B
2	348.0	Measles
3	110.0	BMI
4	1342.0	Polio
5	750.0	Diphtheria
6	739.0	Incidents_HIV
7	5.0	GDP_per_capita
8	7.0	Thinness_ten_nineteen_years
9	1975.0	Thinness_five_nine_years
10	1708.0	Schooling

	VIF Factor	features
0	24262.0	Alcohol_consumption
1	1459.0	Hepatitis_B
2	132.0	Measles
3	927.0	BMI
4	23235.0	Polio
5	5289.0	Diphtheria
6	6256.0	Incidents_HIV
7	911.0	GDP_per_capita
8	953.0	Thinness_ten_nineteen_years
9	683.0	Thinness_five_nine_years
10	978.0	Schooling

	VIF Factor	features
0	5383.0	Alcohol_consumption
1	35.0	Hepatitis_B
2	219.0	Measles
3	21.0	BMI
4	6605.0	Polio
5	691.0	Diphtheria
6	830.0	Incidents_HIV
7	18.0	GDP_per_capita
8	17.0	Thinness_ten_nineteen_years
9	1425.0	Thinness_five_nine_years
10	1473.0	Schooling